**** BEGIN LOGGING AT Tue Apr 04 03:00:02 2017 Apr 04 07:03:54 Do we support setting USE_NLS="no" generally? Even musl enables NLS, It's not documented anywhere and half a dozen recipes seem to fail with it... Apr 04 07:23:36 jku: we set that for -native don't we? Apr 04 07:29:47 RP: yeah we do Apr 04 07:30:58 jku: so we do at least partially support it :) Apr 04 07:37:44 Hi! Has anyone else seen false positive QA errors from out of tree kernel modules in Jethro: QA Issue: *kernel-module* rdepends on *-yocto-standard, but it isn't a build dependency? [build-deps] Apr 04 07:38:25 RP : sure , I just got a bug about net-tools failing to build if USE_NLS is set, realized 6 other recipes fail as well, and noticed there's no documentation on who USE_NLS is meant for (like why is it not a dsitro features for example)... Apr 04 07:38:38 this isn't always reproducible and I can't see patches fixing similar things in poky master Apr 04 07:38:53 RP: ... and am wondering which of these things should have bugs in bugzilla for them? Apr 04 07:38:53 jku: probably because it predates distro_features ;-) Apr 04 07:39:07 aha :) Apr 04 07:39:40 jku: I suspect nobody has realised it would make sense to do that (which it would) Apr 04 07:39:57 jku: fits in with the "native distro_features" issue Apr 04 07:40:14 RP, oh which reminds me... I think I need help with that Apr 04 07:40:36 jku: I saw your question late last night. Happy to try and help, or rburton may be able to Apr 04 07:43:26 RP: ok, I'll finish some other things and try bothering one of you later today Apr 04 09:07:38 rburton: "get a patch on the list asap as 2.3 is freezing" re commercial flag on gst-omx. Sorry, I got carried away. Which list is that? Apr 04 09:11:34 phako[m]: if you're sending a patch for gstreamer1.0-omx then openembedded-core@lists.openembedded.org Apr 04 09:20:42 ah. thanks. I just found out that libomxil is also marked as commercial, though Apr 04 09:25:30 i wonder why, surely omxil is just a wrapper around hardware stuff Apr 04 09:25:32 oh wow, bellagio is stone-old Apr 04 09:28:13 ah Apr 04 09:28:31 I suspect something. bellagio uses ffmpeg to implement the actual codecs Apr 04 09:28:54 those could cause the "commerciality" Apr 04 09:35:34 does anyone actually use bellagio? that looks stone-old Apr 04 09:40:03 otavio: do you guys still use the omx stuff? Apr 04 09:41:58 well otherwise it doesn't kill us to just enable the commercial fiag for gst-omx, I was mainly just wondering. Apr 04 12:10:00 * gtristan unsubscribes from high traffic list again Apr 04 12:10:21 But thanks to RP and rburton for taking care of jpeg turbo issues ! :) Apr 04 12:13:47 gtristan: i think you can set it as post-only Apr 04 12:15:53 yeah I remember you said that, and couldnt find it in the subscription page, maybe its in a user preferences or smth Apr 04 12:16:21 * gtristan will indulge and take a look Apr 04 12:20:29 gtristan: http://lists.openembedded.org/mailman/listinfo/openembedded-core - bottom bit, you need your email/address+password Apr 04 12:58:18 RP, "Mail Delivery" check disabled basically ? Apr 04 12:58:45 * gtristan has it setup this way now, I guess it means I can post but will only receive what I am on CC for Apr 04 12:59:39 gtristan: right Apr 04 13:01:54 rburton: omx? Apr 04 13:01:58 rburton: or imx? Apr 04 13:02:43 We are working on a systemd based system and one thing which is confusing us is the volatile handling; it seems to not be happening Apr 04 13:25:23 I'm working with otavio in a systemd based system and we see that openssh installs /etc/default/volatiles/99_sshd. This file is used to creates /var/run/sshd directory and /var/log/lastlog file, but this is not happen. If I access system using SSH I see message lastlog_openseek: Couldn't stat /var/log/lastlog: No such file or directory. I see that systemd uses files in /etc/tmpfiles.d/ to creates volatiles files and dirs, so if I move 99_sshd file to Apr 04 13:25:23 /etc/tmpfiles.d/ systemd creates /var/log/lastlog on boot. So, we need to move all files from /etc/default/volatiles to /etc/tmpfiles.d or is there something that is not creating volatiles dirs? Apr 04 13:37:38 Does your system support lastlog? Apr 04 13:37:56 otherwise, PrintLastLog=no Apr 04 13:43:26 Also, yes, I think etc/tmpfiles.d is the systemd way of /etc/default/volatiles Apr 04 13:49:31 phako[m]: lastlog is just an example, this also happens with wpa_supplicant, there's a /etc/default/volatiles/99_wpa_supplicant file but this file doesn't creates /var/run/wpa_supplicant dir. Apr 04 14:10:01 /etc/default/volatiles is from the old populate-volatile.sh init script, I think you need to transitiont o /etc/tmpfiles.d for systemd. Though there seems to be a slight inconsistency Apr 04 14:55:56 YPTM: Ready-Access Number: 8007302996 Access Code: 2705751 Apr 04 14:55:56 YPTM - armin is on Apr 04 14:57:13 YPTM: Stephen Joined Apr 04 14:59:07 YPTM: Richard joined Apr 04 14:59:09 YPTM: Stephano joined Apr 04 14:59:25 YPTM: Joshua joined Apr 04 14:59:41 YPTM: I'm on (Mark) Apr 04 15:02:03 YPTM: Saul here Apr 04 15:03:07 YPTM: Randy joined. Apr 04 15:03:41 YPTM ross joined Apr 04 15:07:21 phako[m]: indeed; we will be adding patches for the services we need Apr 04 15:07:46 berton: yes, every temporary path needs to be both in volatiles for sysvinit and tmpfiles for systemd at this time. Apr 04 15:08:38 YPTM: tlwoerner joined Apr 04 15:08:45 YPTM: Svein silently joined in the background Apr 04 15:13:17 otavio: actually, that explains the ugly hack in our system I discovered last week Apr 04 15:14:23 lol Apr 04 15:23:22 https://wiki.yoctoproject.org/wiki/Newcomers Apr 04 15:25:04 * rburton trying to remember what the word gnome uses is Apr 04 15:25:48 turns out, it's Newcomers Apr 04 15:25:49 :) Apr 04 15:26:02 yeah, used to be Gnome Love Apr 04 15:26:32 OK, I put it in the minutes as NEWCOMERS. Apr 04 15:27:23 I blatantly stole that term from Gnome :) Apr 04 15:27:28 "borrowed" Apr 04 15:27:33 inspired Apr 04 15:27:40 exactly Apr 04 15:28:49 YPTM is over. Apr 04 15:30:21 I am a big proponent for having a system for doing the layer setup/fetching. A configuration management system. Apr 04 15:31:24 Since there is no such system in yocto, then each project and each implementator is left on their own for this Apr 04 15:31:54 And this is what we see from e.g. the different SoC vendors, having their own system on top of yocto Apr 04 15:32:06 My 2 cents... Apr 04 15:39:14 sveinse: i'm quite sure that one day the yocto project will have such a tool, there certainly is a lot of interest in such a tool Apr 04 15:42:12 RP: https://www.amazon.ca/Project-Reference-Manual-Richard-Purdie/dp/9888381989/ref=sr_1_3?ie=UTF8&qid=1491320467&sr=8-3&keywords=yocto Apr 04 15:42:44 the tough part is people have pretty strong opinions about such tools, and not all tools necessarily meet all needs. i think everyone agrees it'd be nice to settle on something, but it's non-trivial. the new layersetup tool for example uses repo under the hood, but requires use of a layer index, which doesnt' work for everyone, and not everyone likes repo Apr 04 15:43:37 kergoth: nicely put Apr 04 15:44:22 stephano: more books for the newcomers page? https://www.amazon.ca/s/ref=nb_sb_noss?url=search-alias%3Daps&field-keywords=yocto Apr 04 15:48:06 tlwoerner: thanks! Apr 04 15:48:47 hmmm, Yocto Project Reference Manual by Richard Purdie, sound sketchy. Apr 04 15:49:13 stephano: yea... i was wondering if RP was aware of it Apr 04 15:49:21 :) Apr 04 15:49:45 packt wants the corner the market on yocto books :-) Apr 04 15:52:46 stephano: oh... and there's a printed bitbake manual too by Richard as well, sadly those manuals change weekly, any printed form would be out of date before its delivered Apr 04 15:53:10 tlwoerner: yeah, strange magic there Apr 04 15:56:57 stephano: you alluded to something in the meeting that i think requires more thought... when people use google to find information wrt OE/yocto, because of google's page ranking algorithm, they are more likely to find older versions of the manual than newer ones Apr 04 15:58:20 or they'll find old blog posts whose instructions are very out of date, or tutorials that have the users trying to perform dylan builds (for example) Apr 04 16:00:17 I can confirm that. My google puts the Yocto manual 1.8 on top, so I always have to manually edit the URL to use a later one or latest Apr 04 16:00:32 is there a way to update the manuals that are hosted on the oe/yocto websites such that a floating window with bold, red lettering will follow the user around saying something to the effect of "you are reading an older version of this manual" Apr 04 16:00:52 "are you sure you want old information? here's a link to the latest manual" Apr 04 16:01:29 tlwoerner: Twisted do it like this: http://twistedmatrix.com/documents/8.0.0/api/index.html Apr 04 16:03:41 sveinse: that's not bad. due to the hugeness of the mega-manual and the likelihood of being dropped somewhere in the middle of it, i'd want something that is not just at the top Apr 04 16:03:46 stephano: its a print of the YP manual Apr 04 16:04:04 tlwoerner: that's a good idea (floating warning) Apr 04 16:04:39 tlwoerner: I think someone would need to show scott rifenbark how to do it Apr 04 16:04:41 tlwoerner: I'm no CSS expert, but you can have a float that follows the viewport, right? Apr 04 16:05:38 RP: ah okay Apr 04 16:05:42 sveinse: it sounds like you already know more than me about it ;-) Apr 04 16:06:09 tlwoerner: Agreed. I'd even settle for a dropdown box that lets the user choose the version, and then call out the fact that they are not looking at the latest. Apr 04 16:06:09 RP: were you aware of this print? or is this some random person/group printing manuals they find on the internet? Apr 04 16:06:25 tlwoerner: the latter Apr 04 16:18:29 tlwoerner ya, I'd guessed it's someone snagging 'free' PDFs and publishing them Apr 04 16:29:00 Pardon for my extremely crude hack, but this is principally one way of doing the fixed text on top on the mans: https://embed.plnkr.co/56fWbQd1X9DZXbSlOUSM/?show=preview -- it need hugely more finesse of course Apr 04 16:32:47 ^ tlwoerner Apr 04 16:49:05 zeddii_home: Is there a way for a recipe to discover what KERNEL_VERSION is actually being built? I know the KERNEL_VERSION variable exists, but not sure if it's available where I want it (it's an "export" variable) Apr 04 17:02:54 sgw_ what is the purpose of the recipe? userspace tool (kernel independent or kernel module or? Apr 04 17:03:07 independent userspace tools can check the headers for capabilities Apr 04 17:03:22 modules, there are rules for out of kernel modules which can do similar header checks for versions Apr 04 17:03:27 (just different headers being checked) Apr 04 17:04:13 (that is the only generic way to do it, since anyone can put in their own kernel with their own rules and even throw away all of the YP kernel tooling) Apr 04 17:04:39 fray: unfortunately, an inter-dependency on kernel version for a patch to cryptodev-module, a 4.9 kernel has a backported 4.10 patch that changes the API that cryptodev-module uses and knows about Apr 04 17:05:12 fray: so I have to force the cryptodev patch based on what kernel and version of kernel Apr 04 17:05:57 module is a kernel module right? Apr 04 17:06:25 so you can depend on the virtual/kernel -- write a small chunk (or use the same chunk from the YP code) to inspect the kernel sources for the version (prior to or in the do patch) Apr 04 17:07:15 fray, correct, but the problem is the kernel version still reports 4.9 which is the old API for cryptodev, but it has the 4.10 api via backport, yes really screwed things up Apr 04 17:07:52 if you can't discover teh backport by other means then I'd say you are stuck and it's implementation specific at that point and no longer generic Apr 04 17:08:04 fray: maybe a python to inspect the git history (really hate that hack) Apr 04 17:08:11 can you look for ioctls or anything else like that? Apr 04 17:08:47 it's a change to kernel mm function call Apr 04 17:08:49 perhaps something as simple as a PKGCONFIG or other override switch for the developer to just set the API -- otherwise use the version check via headers Apr 04 17:09:32 if it's something that is in a header you can do the 'autoconf like' approach of checking the API available Apr 04 17:09:42 (cpp and all that looking for a compile problem) Apr 04 17:10:26 git commit check hack would be simpler! Apr 04 17:10:39 but not necessarily portable (if that is your goal) Apr 04 17:17:01 fray: may not need to be portable, it's will be within a layer and will only check for when virtual/kernel is the kernel I need to check, so known git history Apr 04 17:29:23 kanavin: any luck on the gpg stuff? Apr 04 18:20:02 alimon: are you currently working on yocto-compat-layer.py? I've started to use it against different BSP layers and the current output isn't useful enough for me. Instead of doing lots of manual work, I find it easier to enhance the tool. But I don't want to risk conflicts with changes that you might be doing right now. Apr 04 18:38:00 hi, I am building Krogoth with b2qt and in my local.conf I enabled ENABLE_QTQUICKCOMPILER = "1" Apr 04 18:38:40 I have a commercial qt license but yocto fails with ssh: connect to host codereview.qt-project.org port 22: Connection refused Apr 04 18:38:40 fatal: Could not read from remote repository. Apr 04 18:38:51 Please make sure you have the correct access rights Apr 04 18:38:51 and the repository exists. Apr 04 18:39:38 how do I setup the credentials? couldn’t find any documentation regarding this Apr 04 19:07:03 HavoK_: it's in the qt wiki Apr 04 19:07:11 and the compiler is open source now Apr 04 19:08:04 pohly: i'm not working right know, i acked some patches from Mark Hatle in the ML Apr 04 19:08:07 or so I thought, but it isn't really needed with 5.8 and is going away anyways Apr 04 19:09:11 pohly: i have other patch for the bitbake-diffsigs stuff but it requires some patches from paul Apr 04 19:09:24 HavoK_: https://wiki.qt.io/Setting_up_Gerrit Apr 04 19:10:02 make sure you use the right port Apr 04 19:14:19 alimon: which patches do you need from Paul? Can you share your wip patch? Apr 04 19:18:10 pohly: the first patch related to test_signatures, http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=alimon/yp_compatible Apr 04 19:18:30 and the patches needed are in paul branch http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=paule/sigstuff Apr 04 19:18:42 he added an option to search the diff signature by recipe Apr 04 19:18:50 and improved bitbake-diffsigs Apr 04 19:20:20 alimon: are Paul's patches going to land in 2.3? Apr 04 19:21:51 Alternatively the yocto-compat-layer.py could just search for the sigdata itself. Doing that based on recipe and hash isn't hard. Apr 04 19:22:54 pohly: yes paul knows about the requirement and he will push into 2.3 Apr 04 19:23:17 pohly: but the bitbake-diffsigs has that kind of logic to search the sigdata file Apr 04 19:23:31 so i prefer to use the tool instead of duplicate the logic Apr 04 19:23:42 alimon: I agree that it's better to do it there, I was just considering a plan B. Apr 04 19:24:08 pohly: good Apr 04 19:25:56 @sandsmark thanks, I tried that but didn’t seem to work, I will try it again maybe I missed something Apr 04 19:26:58 make sure SSH isn't blocked Apr 04 19:27:02 on your end Apr 04 19:28:12 I prefer the yocto-compat script just to print out that something was changed with the recipe/task hashA/hashB style.. Apr 04 19:28:19 I'm capable of runnign bitbake-diffsigs myself.. Apr 04 19:28:26 (when it runs it, it is REALLY slow) Apr 04 19:47:06 @sandsmark ssh shouldnt be blocked I’m using it with bitbucket. I just set it all up. Is there a way I can test my access? Apr 04 19:47:40 I can’t seems to find codereview.qt-project.org/qt/tqtc-qmlcompiler Apr 04 19:47:43 online Apr 04 20:30:36 | Running edk2 build for OvmfPkgX64 Apr 04 20:30:36 | Usage: build.exe [options] [all|fds|genc|genmake|clean|cleanall|cleanlib|modules|libraries|run] Apr 04 20:30:36 | Apr 04 20:30:36 | build.exe: error: Option -n only allows one instance in command line! Apr 04 20:30:39 anyone run into this with ovmf? Apr 04 20:36:23 HavoK_: iirc they announced that the compiler would be released as open source with qt 5.8 so the paths and stuff might have changed, but I can't find it Apr 04 20:36:40 but if you're feeling experimental you can test qmlcachegen instead, which serves kind of the same purpose Apr 04 20:38:34 hmm… I wonder if I can find the new links in a newer version of yocto Apr 04 20:38:55 how do i even use qmlcachegen. can’t see any documentation for it. Apr 04 20:41:09 it's coming in qt 5.9 Apr 04 20:41:53 with 5.8 the compilation result of qml files are cached to disk, with qmlcachegen you can generate them beforehand using basically the same code Apr 04 20:42:26 but yeah, I'd contact your FAE or whatever at the qt company for correct links to the compiler Apr 04 20:46:12 Thanks! Apr 05 02:10:04 i've been having network issues lately so it's probably just me... but does anyone have trouble reaching *yoctoproject.org ? Apr 05 02:53:01 tlwoerner: seems to be working at the moment as far as I can tell **** ENDING LOGGING AT Wed Apr 05 03:00:03 2017